Research ResourceImmunology

Comprehensive Expression Profiles of Genes for Protein Tyrosine Phosphatases in Immune Cells

See allHide authors and affiliations

Science Signaling  31 Aug 2010:
Vol. 3, Issue 137, pp. rs1
DOI: 10.1126/scisignal.2000966


The phosphorylation and dephosphorylation of signaling molecules play a crucial role in various cellular processes, including immune responses. To date, the global expression profile of protein tyrosine phosphatases (PTPs) in various immune cells has not been described. With the RefDIC (Reference Genomics Database of Immune Cells) database compiled by RIKEN (Rikagaku Kenkyusho), we examined the expression patterns of PTP-encoding genes in mice and identified between 57 and 64 PTP-encoding genes (depending on cutoff values) that were commonly expressed in immune cells. Cells of different lineages contained additional, unique PTP-encoding genes, which resulted in a total of 58 to 76 genes. Compared with cells from nonimmune tissues, immune cells exhibited enhanced expression of the genes encoding 8 PTP-encoding genes, including Ptprc, Ptpn6, and Ptpn22, but had barely detectable expression of 11 PTP-encoding genes, including Ptprd and Tns1. Each immune cell lineage had between 2 and 18 PTP-encoding genes expressed at relatively high or low extents relative to the average expression among immune cells; for example, Ptprj in B cells, Dusp3 in macrophages, Ptpro in dendritic cells, and Ptprg in mast cells. These PTPs potentially play important roles in each cell lineage, and our analysis provides insight for future functional studies.


Protein tyrosine phosphorylation is a fundamental mechanism for numerous physiologically important phenomena, such as cellular activation, proliferation, differentiation, migration, homeostasis, and death, and is strictly controlled by protein tyrosine kinases (PTKs) and protein tyrosine phosphatases (PTPs) (1, 2). Protein phosphorylation is involved in various diseases, including cancer and immune-related disorders. In 1988, the leukocyte common antigen CD45 was discovered to act as a PTP (3, 4). In parallel, CD45 was reported to regulate T cell receptor (TCR) and B cell receptor (BCR) signaling (5), and these findings converged into a demonstration that PTPs were critical for the regulation of immune cells (6, 7, 8). Since then, many researchers have attempted to identify previously uncharacterized PTPs in humans and mice through molecular biology methods, such as the screening of complementary DNA (cDNA) libraries with probes (9) and degenerative polymerase chain reaction (PCR) with a conservative sequence surrounding the motif VHCSAGxGR[T/S]G, which is found in the catalytic domains of classical PTPs such as PTP1B and CD45 (10). Consequently, numerous PTPs were rapidly identified and their expression patterns and functions were individually investigated. Until completion of the sequencing of genomes, however, the number of PTPs that remained to be discovered was unclear.

Subsequent to an overview of about 38 classical PTPs (11), Alonso et al. expanded the PTP family and created a new comprehensive catalog of PTPs, which contained 107 human and 106 mouse PTPs (12). These 107 human PTPs are close in number to the 90 PTKs that have been previously cataloged (13, 14). Not including those enzymes that are apparently inactive, 85 PTKs and 81 PTPs are strictly tyrosine-specific, which suggests comparable levels of complexity and substrate specificity among these enzymes. PTKs and PTPs often have opposing effects on a single substrate while also recognizing different substrates in an overlapping manner. In many cases, PTKs and PTPs that are closely located to each other in their intracellular localization do not necessarily have the same protein interaction domains, making it difficult to predict their networks of interactions. The 107 human PTPs can be classified into four categories (classes I, II, and III Cys-based and Asp-based) with several subgroups on the basis of their structures and substrate specificities (12). The largest class I Cys-based group contains 21 transmembrane, classical receptor-like PTPs (RPTPs), 17 intracellular, non–receptor-type PTPs (NRPTPs), 11 mitogen-activated protein kinase (MAPK) phosphatases (MKPs), 19 atypical dual-specificity phosphatases (DSPs), 3 Slingshots, 3 PRLs (proteins of regenerating liver), 4 CDC14s (cell division cycle 14s), 5 PTENs (phosphatase and tensin homologs deleted from chromosome 10), and 16 myotubularins.

Since the first report that mutations in MTM1 cause X-linked myotubular myopathy in humans (15), the involvement of many PTPs in diseases has been identified. Mutation in EPM2A leads to progressive myoclonus epilepsy (16). In immune cells, mutations in PTPRC cause immunodeficiency and autoimmune diseases, depending on their position (17, 18). A single-nucleotide polymorphism (SNP) in PTPN22 is associated with type 1 diabetes (19, 20), and many other groups have made similar observations in various autoimmune diseases, indicating the pathophysiological importance of PTPN22 in immune regulation (21, 22). Accumulating evidence shows that SNPs, as well as conventional mutations in PTPs, are important risk factors for many diseases, including cancer (2, 23, 24). Moreover, a comprehensive analysis of SNPs in the human genome is now in progress and is expected in the near future to reveal a large number of SNPs that confer disease vulnerability (25).

Several open-access databases are available, such as the Gene Expression Omnibus from the National Institutes of Health, ArrayExpress from the European Molecular Biology Laboratory, and SymAtlas from the Novartis Research Foundation (26). Despite the progress that has been made, supported by genetics and bioinformatics, no report describing the expression patterns of all of the PTP family members has been made. Such a report could enable the PTP transcriptome (PTPome) of a particular cell lineage to be determined and the relative importance of the physiological roles of PTPs in particular immunological phenomena to be estimated. Hijikata et al. at Rikagaku Kenkyusho (RIKEN) in Japan have also developed an open-access database named the Reference Genomics Database of Immune Cells (RefDIC), which enables raw data to be downloaded in its entirety (27). More than 400 cell types or tissue samples are now available in this database, and this number continues to increase gradually. We downloaded data related to PTPs from this database and analyzed the resulting information. Here, we provide information about the PTPome of many immune cells, which should be beneficial to researchers studying immunology, signal transduction, and inhibitor design.


Overview of PTP-encoding genes in immune cells

Using the open-access database RefDIC, we first generated a heat map to visualize the relative extent of expression in various immune cells of all of the genes that encode murine PTPs (Fig. 1). The immune cells shown include T cells (naïve, resting CD4+ T cells), natural killer (NK) cells, NKT cells, B cells, dendritic cells (DCs), peritoneal macrophages, mast cells, and neutrophils. A nonimmune tissue (brain) was also included as a control. At first glance, the overall expression patterns seem to be largely similar among all the immune cells, indicating that these cells are closely related to each other. We noticed that genes encoding classical RPTPs were poorly expressed in immune cells relative to brain, whereas those genes that encode classical NRPTPs and MKPs were relatively strongly expressed. Moreover, genes that encode PRLs and most of the members of the myotubularin family were prominently expressed, although their functional importance has only begun to be investigated. In contrast, the control brain tissue expressed a broad spectrum of genes encoding PTPs, including RPTPs.

Fig. 1

Overall view of the expression of PTP-encoding genes in immune cells. A heat map of all murine PTP-encoding genes in immune cells and a nonimmune tissue is shown. The immune cells consisted of T cells (naïve, resting CD4+ T cells), NKT cells, NK cells, splenic B cells, bone marrow–derived immature DCs (iDCs), peritoneal macrophages, bone marrow–derived mast cells, and neutrophils. A nonimmune tissue (brain) was also included as a control. At the bottom, several genes for lineage marker molecules such as CD3e (for T cells), NK1.1 (for NK and NKT cells), CD19 (for B cells), CD11c (for DCs), c-kit (for mast cells), and Gr1 (for neutrophils) are included to confirm the reliability and purity of each lineage sample. The abundances of PTP mRNAs are represented by the relative values in log2 space. Multiple probes were used to detect most of the PTPs, with occasionally conflicting results. The reliability of the probes is shown on the RIKEN Research Center for Allergy and Immunology RefDIC site ( The hybridization intensity log2 values are represented by the color gradations at the bottom.

Immune cell–specific PTPs

To identify PTPs specific to immune cells, we calculated the mean relative expression patterns of PTP-encoding genes of the eight immune cell types described earlier, as well as of five nonimmune tissues. The average scores of the nonimmune cells were subtracted from those of the immune cells. The nonimmune tissues consisted of kidney, liver, brain, muscle, and heart. Some PTPs exhibited at least a fourfold difference (in either direction) in the extent of their expression in the immune cells compared to that in the nonimmune cells (difference in log2 >2.0) (Table 1). The group of genes with the most enhanced expression encoded many classical NRPTPs, such as Ptpn6, Ptpn22, Ptpn18 (>8.0-fold), Ptpn1 (>5.6-fold), and Ptpn2 (>4.0-fold). In particular, the top three PTPs (Ptprc, Ptpn6, and Ptpn22) are among the most intensively investigated PTPs. Mutations in the genes encoding these PTPs in humans or knockout (KO) mice result in immune-related abnormalities, indicating that these PTPs play critical roles in immune regulation. For example, mutations in human PTPRC result in severe combined immunodeficiency disease or multiple sclerosis (17, 18), and the reduced expression of human Ptpn6 possibly causes leukemia (28). Ptpn2-deficient mice exhibit specific defects in bone marrow, B cell lymphopoiesis, and erythropoiesis, as well as impaired T and B cell functions (29). Mice heterozygous for Ptpn2 on a Ptpn1-deficient background (Ptpn2+/−Ptpn1−/−) develop signs of inflammation (30).

Table 1

PTP-encoding genes specific to immune cells. PTP-encoding genes with relatively high or low expression in immune cells were determined by subtracting the average expression scores in the nonimmune tissues from those of the immune cells. The PTP-encoding genes were ranked on the basis of the difference in log2 hybridization intensity values. For each gene, the protein that it encodes is shown in parentheses.

View this table:

In contrast, the group of PTP-encoding genes that were least expressed in immune cells contained many RPTPs, such as Ptprd, Ptprb, Ptprk, Ptprm, Ptprf, and Ptprg (all <8.0-fold lower in expression relative to that in nonimmune cells), as well as Ptprn and Ptprs (both <4.0-fold lower in expression). Many atypical DSPs, such as Dusp3 (<8.0-fold lower); Dusp14, Dusp27, and Dusp26 (<5.6-fold lower); and Dusp28 (<4.0-fold lower), were also included in the latter group. These observations are consistent with earlier results (Fig. 1). The expression data shown here represent basal expression in each cell lineage because they were collected from naïve or untreated cells and not from previously activated or differentiated cells. Thus, we cannot rule out the possibility that those PTP-encoding genes whose expression is barely detectable in untreated cells might play some role in the immune system and that other PTPs might also appear after cellular activation or differentiation in various immune responses not depicted here.

PTP-encoding genes common and specific to each cell lineage of the immune system

Next, we attempted to determine the entire PTP transcriptome, the so-called PTPome, of each cell lineage. When the cutoff value of hybridization intensity for the expression of PTP-encoding genes was set to log2 >3.5, the number of PTP-encoding genes commonly expressed in all eight immune cell lineages (hereafter called “common PTPs”) was 57 (Fig. 2). When a cutoff value of log2 >3.0 was used, 64 PTPs were included. Moreover, CD4+ T cells had 3 (log2 >3.5) or 8 (log2 >3.0) more PTP-encoding genes in addition to the common genes, resulting in a total of 60 (log2 >3.5) or 72 (log2 >3.0) PTP-encoding genes from the 100 genes on the microarray that was used. In the same way, B cells had 1 (log2 >3.5) or 5 (log2 >3.0) more PTP-encoding genes in addition to the common genes, resulting in a total of 58 (log2 >3.5) or 69 (log2 >3.0) PTPs, whereas mast cells had 7 (log2 >3.5) or 11 (log2 >3.0) more PTPs, resulting in a total of 64 (log2 >3.5) or 75 (log2 >3.0) PTPs. These numbers largely support the observation that ~67% of the genes in the entire genome can be assigned a score that represents substantial expression in at least one immune cell type (31). To see the entire picture, we again used the expression data for naïve or untreated cells of each lineage, which represented the basal extent of expression of PTP-encoding genes. Tanzola and Kersh previously reported the MKP and DSP transcriptomes in the mouse thymus through Northern blotting analysis, in which the genes for 7 MKPs and 3 DSPs, but not the 16 other MKPs and DSPs, were expressed (32). Among the 11 MKP- and 15 DSP-encoding genes that the authors tested, genes for 10 MKPs and 10 DSPs were consistent with our results for resting, naïve CD4+ T cells. Regarding the differences between the results of Tanzola and Kersh and ours, we are uncertain whether these discrepancies can be ascribed to differences between thymocytes and peripheral T cells or to the different methods that were used (that is, microarray and Northern blotting analyses).

Fig. 2

Expression of PTP-encoding genes common and specific to each cell lineage of the immune system. The average expression values of the PTP-encoding genes were calculated for each cell population, as shown in Fig. 1. The expression thresholds were set as log2 >3.5 (in blue) or 3.0 (in brown). The PTP-encoding genes expressed in all eight types of cells are listed as common PTPs in a center frame, whereas those genes expressed in only particular cell types are shown in the frames for each cell type. Thus, each cell population has PTP-encoding genes in its own frame, as well as in the common center frame. The dotted lines separate subgroups such as RPTP, NRPTP, MKP, and others, as shown in Fig. 1. The asterisk (*) indicates PTP-encoding genes that exhibited different extent of expression with multiple probes. Ptpn7 (HePTP) is not found in the RefDIC database but is highly expressed in immune cells according to ImmGen, the Web site of the Immunological Genome Project Consortium (

Highly or marginally expressed PTPs in each cell lineage

We further tried to identify unique or characteristic PTP-encoding genes, whose expression was higher or lower in each cell lineage, that might confer lineage-specific properties by subtracting the average scores of all eight immune cell types from the scores of each lineage. The PTP-encoding genes most highly expressed in given cell types were Dusp10 in CD4+ T cells, Dusp2 and Eya2 in NKT and NK cells, Ptprj in B cells, Dusp3 in macrophages, Ptpro in immature DCs, Ptprg in mast cells, and Cdc14a and Dusp1 in neutrophils (Fig. 3). Dusp10-deficient T cells, as well as innate immune cells, reportedly produce substantially increased amounts of cytokines relative to wild-type cells (33). Ptprj has a stimulatory function in B cells and macrophages (34). Dusp2 is much more highly expressed in lymphoid cells than in myeloid cells. We also analyzed lineage marker genes, such as Cd3e and Cd19, which are usually highly cell-specific. Although most PTP-encoding genes were less specific than lineage markers to a particular lineage, the expression of Ptprg in mast cells is more prominent than that of the mast cell lineage marker gene Kit.

Fig. 3

Lists of highly or marginally expressed PTP-encoding genes in each cell lineage. Lineage-specific PTP-encoding genes were determined by subtracting the average expression values of all eight different immune cells from those in each cell lineage. The genes were then ranked on the basis of the difference in these log2 hybridization intensity values. PTP-encoding genes with a difference in log2 of more than 2.0 are shown. Positive values (in red), negative values (in green), and lineage marker molecules (in blue) are also shown. In cases in which two or more probes were available for hybridization, two intensity values or a value range is shown. Cd3e, a lineage-specific gene in T cells; Cd19, a lineage-specific gene in B cells; Klrb1c, a gene that encodes NK1.1 in NK and NKT cells; Itgax, a gene that encodes CD11c in DCs; Kit, a gene that encodes c-kit in mast cells; Ly6c1, a gene that encodes Ly-6C, which is a part of Gr1 in neutrophils and macrophages.

PTP-encoding genes in DCs and macrophages

Bone marrow–derived immature DCs and peritoneal macrophages share 6 PTP-encoding genes (Ptpro, Dusp3, Ssh3, Dusp26, Epm2a, and Mtmr7) in addition to the 64 common PTP-encoding genes (log2 >3.0) (Fig. 2); however, only 2 PTP-encoding genes (Ptpro and Sbf2) are equivalently expressed in these cell types (Fig. 3), indicating that although the existing PTP members are similar, the extent of their expression is different. Upon stimulation of DCs and macrophages with bacterial lipopolysaccharide (LPS) for 4 or 48 hours, a number of PTP-encoding genes undergo alterations in their expression (Fig. 4). Ptprj and Mtmr7, as well as Cd40 and Cd86, which encode well-known activation markers, were commonly increased in expression, whereas the expression of Dusp7 and Cdc25b decreased. In DCs, the expression of an additional five PTP-encoding genes (Dusp14, Dusp4, Dusp2, Dusp16, and Ptpn2) was increased after 4 hours of LPS treatment, whereas four PTP-encoding genes (Ptprf, Ssh1, Mtmr1, and Ptpn4) were increased in expression at 48 hours, indicating that the genes whose expressions were increased during the early phase included many that encode MKPs and DSPs. This finding is consistent with the fact that many of the genes that encode MKPs and DSPs were originally identified as immediate-early genes in the cellular response (35, 36). Compared with the basal expression of PTP-encoding genes (Fig. 2), only Dusp14 newly appeared upon stimulation. On the other hand, the expression of many down-regulated PTP-encoding genes in DCs did not overlap between 4 hours (Ssh2, Ptprs, Mtmr2, Sbf1, Cdc25a, and Tns1) and 48 hours (Ptpn22, Cdkn3, Ptpn18, Cdc25c, and Ptpro). This observation suggests that early changes in gene expression were directly induced by LPS, but that later cellular responses might be mediated by cytokines produced during the early phase, resulting in the activation of different signaling pathways, which lead to alterations in the expression of a distinct set of PTP-encoding genes.

Fig. 4

Expression of PTP-encoding genes in DCs and macrophages. Alterations in the expression of PTP-encoding genes in bone marrow–derived DCs and peritoneal macrophages (Mϕ) after stimulation with bacterial LPS for 4 and 48 hours are shown. PTP-encoding genes with a difference in log2 of more than 2.0 are shown. Positive values (in red), negative values (in green), and lineage marker molecules (in blue) are also shown. In cases in which two or more probes were available for hybridization, two intensity values or a value range is shown. The expression of Cd40 and CD86, which encode markers of activated DCs and macrophages, is increased upon activation.

PTP-encoding genes in T cell subsets

T cells can be divided into a number of subpopulations with distinct roles. To identify possible links between PTPs and a particular role in T cells, we examined the preferential expression of PTP-encoding genes in each T cell subpopulation. Compared with resting, naïve CD4+ T cells, activated CD4+ T cells had four PTP-encoding genes that were increased in expression and eight that were decreased in expression (Fig. 5). Activated CD8+ T cells had three PTP-encoding genes that were increased in expression and three that were decreased in expression compared with resting, naïve CD8+ T cells, which had no alterations in gene expression relative to that of resting naïve CD4+ T cells. Dusp4 and Dusp16 were both commonly increased in expression in activated T cells, whereas Ptpre and Eya2 were commonly decreased. In T helper 1 (TH1) cells, in which Ifng and Tbx21 (which encode TH1-specific markers) were abundantly expressed, Ptprk, Ptprg, Ptprz1, and Dusp3 were preferentially expressed. In particular, Ptprk, Ptprg, and Dusp3 were newly expressed compared with PTP-encoding genes expressed in resting, naïve T cells (Fig. 2). On the other hand, in TH2 cells, in which Il4 and Gata3 were abundantly expressed, Dusp10, Ptpn13, and Dusp6 were preferentially expressed. Data for the recently established T helper lineage, TH17, were not yet available in RefDIC. Ptprk-deficient rats, but not mice, exhibit a developmental defect involving CD4+ single-positive thymocytes, implying a critical role for Ptprk in T cell maturation in the thymus (3739). Regulatory T cells, in which Foxp3, which encodes the lineage-specific transcription factor Foxp3, is abundantly expressed, exhibited increased expression of Dusp4 and Tns1 and decreased expression of Eya2 compared to that in conventional T cells. Intraepithelial CD8+ T cells in the small intestine had four PTP-encoding genes that were increased in expression and eight that were decreased in expression relative to naïve CD8+ T cells. In most cases, the roles of these PTPs in their respective T cell subpopulations remain to be investigated.

Fig. 5

Expression of PTP-encoding genes in T cell subsets. The average expression values of PTP-encoding genes were calculated for each subset. Subset-specific PTP-encoding genes were determined by subtracting the average expression values of naïve, resting CD4+ or CD8+ T cells from those of PTP-encoding genes in each subset. PTPs with a difference in log2 of more than 2.0 are shown. Positive values (in red), negative values (in green), and lineage marker molecules (in blue) are also shown. In cases in which two or more probes were available for hybridization, two intensity values or a value range is shown. IEL, intestinal intraepithelial lymphocyte; Cd69, a gene that encodes a marker of activated T cells; Cd62l, a gene that encodes a marker that is decreased in expression upon activation of T cells; Tbx21, a gene that encodes T-bet, a transcription factor predominant in TH1 cells.

PTP-encoding genes in B cells upon stimulation

Next, we examined alterations in the expression of PTP-encoding genes in splenic B cells upon stimulation with antibody against immunoglobulin M (IgM) or with LPS for 1 or 7 hours (Fig. 6). After 1 hour of stimulation, at which time the activation marker Cd86 was increased in expression, Dusp4, Ptpn23, and Dusp5 were increased in expression compared to that in unstimulated cells, whereas Dusp19, Dusp7, Dusp6, Ptpro, and Ssh1 were decreased in expression. Similar to DCs (Fig. 4), the expression of genes encoding MKP and DSP family members also rapidly changed in the B cells. After 7 hours of stimulation, at which time Cd62l was decreased in expression, consistent with cellular activation, a further seven PTP-encoding genes, including Dusp1 and Cdc25b, were decreased in expression. In summary, B cells are characterized by an increase in the expression of Dusp4 and Dusp14 and a decrease in the expression of Dusp6, Ptpro, and Ssh1 until 7 hours after stimulation. Overall, only the expression of Dusp14 and Dusp3 newly appeared upon stimulation (that is, were undetectable in unstimulated cells, but detectable in stimulated cells) compared with cells under unstimulated conditions (Fig. 2).

Fig. 6

Expression of PTP-encoding genes in B cells upon stimulation. Alterations in the extent of expression of PTP-encoding genes in splenic B cells after stimulation with antibody against IgM or with LPS for 1 and 7 hours were determined. PTP-encoding genes with a difference in log2 of more than 2.0 are shown. Positive values (in red), negative values (in green), and lineage marker molecules (in blue) are also shown. Where available, the genes that encode activation markers, such as Cd86, Cd69 (both increased in expression), and Cd62l (which is decreased in expression), are also shown.


Perspectives of PTP analysis

Regarding the roles of PTPs, several approaches and excellent reviews discuss (i) amino acid sequence and three-dimensional (3D) structure (40, 41); (ii) gene expression and protein abundance (42, 43); (iii) regulatory mechanisms of PTP activity including receptor dimerization, phosphorylation, and oxidation (2, 44); (iv) substrate specificity and signaling (45, 46); and (v) KO mice, human SNPs, and diseases (23, 24, 47). Among these approaches, research on the expression of PTP-encoding genes and on the abundance of protein has been limited. Because no single report has focused on the global expression patterns of PTPs in a range of immune cells, we used DNA microarray data from an open-access database compiled by RIKEN. First, we determined the expression profiles of PTP-encoding genes of specific immune cell types by comparing them with the expression profiles in nonimmune tissues. We then determined the PTP-encoding genes specific to each immune cell lineage. Moreover, we identified the expression patterns of PTP-encoding genes in T cell subsets and determined the changes in the expression of PTP-encoding genes upon stimulation of B cells, DCs, and macrophages. These data provide new insight into the roles of PTPs in various cellular mechanisms. On the RefDIC Web site, new results for many other cell subpopulations, such as stem cells, are now accumulating and are available for analysis. Nevertheless, the sample numbers for some cell types remain insufficient to extract particular trends, and many more samples are needed to enable reliable conclusions. Increasing the amount of data that are available for subpopulations of T cells [such as TH17, TH9, and follicular TH (TFH) cells] and for B cells (such as B-1 cells, marginal zone B cells, and follicular B cells) would be of great use to researchers examining the potentially biased expressions of PTPs.

Highly expressed PTP-encoding genes and their functions

As described above, the top three immune-specific PTP-encoding genes, Ptprc, Ptpn6, and Ptpn22, are well known, and their products have been among the most intensively investigated PTPs (Table 1). Numerous reports have demonstrated that these highly abundant PTPs have important effects on physiological phenomena and play critical roles in immune regulation. Among the immune-specific PTPs, the next target genes for extensive study should be Ptpre, Cdc25b, Dusp5, Rngtt, Cdc14a, Mtmr12, Dusp11, and Eya3. Many of these PTPs have been identified as immune- or hematopoietic-specific PTPs, suggesting that mice transgenic for these PTPs might exhibit advantages in fertility or display notable phenotypes. For example, among the previously reported PTPs, Ptpre-deficient macrophages have an alteration that results in the production of tumor necrosis factor–α (TNF-α) and interleukin-10 (IL-10) after stimulation with LPS (48). Dusp5-transgenic mice exhibit a block in thymocyte development at the double-positive stage (49), whereas Cdc25b-deficient oocytes remain arrested at prophase (50). Other PTPs remain to be investigated in mice.

The PTPs whose genes are expressed to the greatest extent in each particular cell lineage are also intriguing (Fig. 3). Such lineage-specific PTPs should be promptly tested for their functional importance in the corresponding cells. Despite previous cases in which KOs of these genes have resulted in lethal embryos, cell type–restricted conditional PTP KO mice might be valuable. Mice deficient in Ptprj, which is highly expressed in B cells (Fig. 3), indeed exhibit a mild alteration in B cell development (34). On the other hand, functional redundancy and compensation often mask the physiological phenotype in mice deficient for a single gene among a family of genes. Mice doubly deficient in Ptprj and exon 6 of Ptprc exhibit even more marked alterations than do mice deficient in each single gene, and the Src family kinase Lyn is a common substrate for both CD148 and CD45 (34). Similarly, deficiency in Ptpn22 and the CD45 E613R allele also contribute cooperatively to break immunological tolerance (51). In contrast, mice triply deficient in Ptpn3, Ptpn4, and Ptpn13 are largely unaffected in terms of their T cell function (52).

Commonly regulated PTP-encoding genes

The stimulation of multiple immune cells (Figs. 4 to 6) results in the increased expression of several common PTP-encoding genes, including Dusp16 (in CD4+ and CD8+ T cells, DCs, and macrophages), Dusp4 (in CD4+ and CD8+ T cells, B cells, and DCs), Dusp14 (in B cells and DCs), and Dusp5 (in CD8+ T cells and B cells), as well as in the decreased expression of Dusp7 (in CD8+ T cells, B cells, DCs, and macrophages), Ptpre (in CD4+ and CD8+ T cells and B cells), Ssh2 (in CD4+ T cells, B cells, and DCs), for example. These observations suggest the possible existence of a common transcriptional mechanism that regulates the expression of these genes. The former genes are supposedly induced by a negative feedback mechanism after the secretion of cytokines and growth factors, whereas products of the latter genes might function as gatekeepers to modulate the activation threshold of naïve immune cells. Dusp5 is a common target gene of IL-2, IL-7, and IL-15 (53), and human DUSP14 reportedly associates with the costimulatory receptor CD28 in T cells and inhibits its signaling (54).

Regulatory mechanisms for PTP-encoding genes

In addition to the results from the simple analysis of messenger RNA (mRNA) expression described here, several complicated mechanisms also need to be considered, such as the use of alternative promoters or of alternative splicing of mRNA (42); such mechanisms might occasionally result in the failure of mRNAs to be detected in microarrays with a probe that might be spliced out. Furthermore, some reports have indicated that the expressions of multiple PTP-encoding genes, including Ptpn11, Ptpn22, Dusp5, and Dusp6, are under the control of a single microRNA (miRNA), miRNA181a (55). However, the expressions of these genes were not always regulated in a parallel fashion in our study, which suggests that factors in addition to miRNA181a might differentially regulate their expression. Various mutations, SNPs, mRNA stability, and posttranslational modifications, such as phosphorylation, dimerization, and oxidation, are able to affect enzymatic activity or protein abundance without changing the amount of the corresponding mRNA. In addition to the RefDIC data, some reports have attempted to simultaneously compare the abundances of mRNAs and their proteins; in these reports, the abundances of many proteins were not proportional to the amounts of their mRNAs (27, 56, 57). Thus, relative differences in the abundances of mRNAs should be interpreted with caution. Indeed, some PTPs are known to change in abundance at the protein level as well. IA-2 (Ptprn) in rat pancreatic islet β cells is cleaved and degraded by a Ca2+-dependent protease, μ-calpain, upon stimulation with a high glucose concentration, which promotes the secretion of insulin (58). In addition, human PTP-PEST (PTPN12) protein is reduced in abundance upon T cell activation (59), but not by regulating the abundance of its mRNA; it is possible that it is cleaved by caspase 3 (60).

Problems in determining the substrates of PTPs

Since the early 1990s, enzymatically inactive or substrate-trapping mutants of enzymes have been used as valuable tools and have revealed the identities of a number of substrates (35, 61). Barr et al. analyzed the molecular surface of the catalytic domain and enzymatic specificity of 22 classical PTPs, providing new and deep understanding of intrafamily PTP diversity, substrate specificity, and crucial hints for the development of small-molecule inhibitors (40). To identify substrates or to determine substrate specificity, many researchers often use a glutathione S-transferase fusion protein that contains a catalytic domain incubated with cell lysates or candidate substrates, under conditions in which subcellular localization and protein interaction domains are ignored. In vitro enzyme assays often show a gentle, gradient difference in substrate specificity. Consequently, many researchers tend to believe that the substrate specificities of PTPs are less strict than those of PTKs. The protein interaction domain is important for the correct subcellular localization of the PTP, the proper association with its binding partners, and the correct recognition of its substrates (62, 63). Thus, full-length molecules need to be expressed at the correct time (for example, in naïve or activated cells) in an appropriate cell type (for example, immature, mature, or differentiated), as observed in the present report, to achieve a sufficient substrate affinity and to identify specific substrates, because the PTP might never come into contact with some substrate candidates in vivo. Tiganis and Bennett have discussed the criteria for identifying bona fide substrates in more detail (45).

The mammalian genome contains at least 10 genes that encode PTPs that are enzymatically inactive (12), many of which belong to a myotubularin subfamily (64). The distal catalytic domains of RPTPs are also often inactive. In Caenorhabditis elegans zygotes, the pseudotyrosine phosphatases EGG-4 and EGG-5 sequester activated and phosphorylated MBK-2, which is a kinase that promotes the oocyte-to-embryo transition (65). This finding suggests that catalytically inactive PTPs have a function similar to that of active PTPs as negative regulators (66).

Future directions

One catalog that would be of great use to researchers is a description of the global network between PTKs and PTPs connected with their substrates. Indeed, such an attempt has already begun with a bioinformatic, “dry” approach (43). More recently, Breitkreutz et al. reported the global protein kinase and phosphatase interaction network in yeast, in which CDC14 played a central role (67). Similarly, as for each cell lineage of the immune system, a combination of bioinformatics and conventional, “wet” approaches of molecular biology that enable further clarification of substrate specificity is required to obtain a detailed and accurate overall picture of kinase and phosphatase networks.

Materials and Methods

Data mining and analysis

We used an open-access database, RefDIC, compiled by RIKEN (Japan), which contains a collection of multiple mRNA expression patterns based on the Affymetrix GeneChip Mouse Genome 430 2.0 Array (27). RefDIC consists of three parts: (i) quantitative mRNA profiles for human and mouse immune cells and tissues obtained with Affymetrix GeneChip technology; (ii) quantitative protein profiles for mouse immune cells obtained from analyses by 2D polyacrylamide gel electrophoresis followed by image analysis and mass spectrometry; and (iii) various visualization tools to cross-reference the mRNA and protein profiles of immune cells. Protein analyses in RefDIC are still limited in number and are insufficient for examining entire PTP expression patterns. Therefore, we have provided information regarding the mouse global PTP transcriptome in this report. Early in 2009, we downloaded all of the raw data used here for further analysis. RefDIC is periodically updated; thus, data from more than 400 mRNA samples have accumulated. Multiple probes were used to detect the expression profiles of the genes encoding most PTPs, with occasionally conflicting results (table S1). The reliability of the probes used is described in detail at the RefDIC site. Here, probe reliability is briefly explained as follows: A, single gene, matched RefSeq entry; B, single gene, matched GenBank entry but not RefSeq entry; C, single gene, matched UniGene entry only; D, single gene, antisense (possibly designed to the complementary strand of the intended reference gene); E, multiple genes; X, nonqualified. The downloaded data were incorporated into Excel files and subgrouped on the basis of the cell lineage, differentiation stage, or activation status. Their average scores and the differences between subgroups were calculated. A heat map was generated with Java Treeview software.

Nomenclature of PTPs used in this study

The classification and nomenclature of the PTPs shown here conform to those in a previous report (12), with some changes based on the HUGO Gene Nomenclature Committee ( As a result, several PTPs were replaced as follows: PtprzPtprz1 (RPTPζ), Mk-styxStyxl1, Dusp23Ptpmt1 (MOSP), Dusp24Dusp26 (MGC1136), Dusp25Dusp23 (VHZ), Dusp26Dusp28 (VHP), Dusp27Duspd1 (FMDSP), Ptp9q22Ptpdc1, TpipTpte2, TnsTns1, Mtmr5Sbf1, and Mtmr13Sbf2. Synonyms and protein names for each PTP-encoding gene are listed in table S1 (12). Although Alonso et al. originally reported 107 human and 105 (or 106) mouse PTP-encoding genes (12), as they suggested, one (Mtmr10) of two missing PTP-encoding genes in mice is present and available in this study. However, the following PTP-encoding genes were not found in the above-mentioned array: Ptprh, Ptprq, Ptpn7, Dusp13A, Tpte2, and Mtmr15. Thus, data for the remaining 100 mouse orthologs were downloaded and incorporated into an Excel file. The data for each gene contained mRNA expression scores from a total of 272 cell samples, including lymphoid, myeloid, and other types of cells. The data for nonimmune cells, such as epithelial cells, stromal cells, germ cells, stem cells, and transformed cells, were excluded, and only 178 lymphoid and myeloid cell samples were used for calculation of the average values among particular cell groups and differential expression values between cell types. These data contained a total of 232 probes for PTP-encoding genes and were used to screen 178 lymphoid and myeloid cells from a total of 272 samples. Ptpn7 is not found in the RefDIC database, but it appears to be highly expressed in immune cells based on data in ImmGen, the Web site of The Immunological Genome Project Consortium ( (31, 68). For T cell subsets, some samples showed the inappropriate expression of genes encoding lineage-specific and activation markers. These samples apparently affected the average scores of particular subsets. To minimize this inaccuracy and obtain appropriate and statistically significant differences, we chose samples that exhibited the proper expression of these marker genes, which included Tbx21 (T-bet), Ifng (interferon-γ), Gata3 (GATA-3), Il4 (IL-4), Foxp3, Cd69, Sell (CD62L), and Il2 (IL-2).


Acknowledgments: We thank A. Hijikata for technical comments. Funding: This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan. Author contributions: Y.A. designed the project, analyzed the data, and wrote the paper. J.Y. supported the project and discussed the paper. Competing interests: The authors declare that they have no competing interests.

Supplementary Materials

Table S1. Compiled raw data downloaded from RefDIC at RIKEN.

References and Notes

Stay Connected to Science Signaling

Navigate This Article